# USA Outlays > A public ledger of U.S. government spending: contract and grant awards (who got > paid, how much, for what) plus the approvals and funding decisions that > authorized them. 60M+ award records across all 50 states, organized under one > policy taxonomy. Built for spending trends nationwide — contract detail is > extra granularity for verification. ## Data layers (never collapsed) - **Awards** — outlays (paid), reimbursements, and obligations (signed contracts/encumbrances not yet paid; excluded from outlay totals). - **Approvals** — the authorizing layer: legislature budget votes, county commission and city council spending decisions, utility/water/school boards, special districts. ## Taxonomy Nested specific → general tree: award → program → topic → domain → function (L1–L3 curated, L4 programs discovered from data), plus orthogonal facets: sector, level, procurement type, funding mechanism, competition, procurement method, lifecycle, instrument, integrity. Classification is on spending purpose, not agency identity. ## Reading the data honestly - Amounts are **stated**, not verified cash movement. Missing values display as "not disclosed", never zero. - Contract ceilings and IDIQ ceilings are flagged and **excluded from spend totals** — never sum them. - One record per entry: exactly one primary topic per award, one facet row per (award_id, facet). ## Machine access - REST API: `https://usa-outlays.ai/api/v1` — key via `x-api-key` header (or `?api_key=`). Subscriptions at /pricing — or pay per call: **$3.40 for one API call, no account needed**. Buy at /pricing; a single-use token is delivered instantly (one request, 24h expiry). Built for agents that just need an answer. - Ask endpoint: `POST /api/v1/ask` with `{"question": "..."}` — an agent resolves the question to the taxonomy and returns a summary answer, the supporting taxonomy, and a raw data packet. One question = one call. Example: "what is the policing budget of utah". - Data surfaces: curated awards/approvals (/api/v1/awards, /api/v1/approvals), raw as-ingested records (/api/v1/raw/awards), taxonomy tree (/api/v1/topics). - OpenAPI spec: https://usa-outlays.ai/openapi.json - MCP server: https://github.com/qchasan/usa-outlays/tree/main/mcp (`search_awards`, `search_approvals`, `get_vendor`, `topic_tree`, `fulltext_search`, `coverage`) - Capability manifest: https://usa-outlays.ai/.well-known/usa-outlays.json ## Docs - Full agent brief: https://usa-outlays.ai/llms-full.txt - Agent landing page: https://usa-outlays.ai/agents - Human API docs: https://usa-outlays.ai/api-docs - Signup: https://usa-outlays.ai/pricing